Hi, I'm Nolan Gelinas!
I'm an Electrical & Computer Engineer and Software Developer
I'm a Senior at Duke University with a strong passion for front-end development. Throughout my academic journey, I have developed a solid foundation in both hardware and software systems, including programming languages such as Python, JavaScript, and Java. Over the years, I've come to embrace website development as my favorite medium for creating interesting and performant applications due to its flexibility and wide availability.
I am enthusiastic about exploring emerging technologies and I enjoy taking on challenging projects that push the limits of what I'm able to do. My goal is to use my skills and knowledge to make a positive impact on society and to contribute to the advancement of technology. Thank you for visiting my page, and I look forward to connecting with you!
Skills
8.5/10
Git
7.5/10
Docker
7/10
GraphQL
7/10
Figma
6.5/10
Fusion 360
5.5/10
GTKWave
5/10
Blender
5/10
Assembly
10/10
TailwindCSS
9/10
React
9/10
JavaScript
8/10
Java
7.5/10
TypeScript
7/10
Verilog
6/10
Python
7/10
Next.JS
4.5/10
Phoenix
Spoken Arabic Digit Classification
This project's goal is to classify digitized spoken Arabic digits using Gaussian mixture modelling. I was given training and testing datasets and use probabilistic modeling to create clusters of the given 13-dimensional data. There are a variety of hyperparameters to choose that improved my modeling accuracy, such as covariance constraints, number of expected clusters per digit, and a clustering algorithm. Once a set of clusters are created for each digit, I use maximum likelihood classification to test my model performance. This involves taking the unknown input digit and comparing the data points to each of the ten Gaussian mixture models. Overall, I achieved over 92% accuracy, which is much higher than the expected 10% using a random guess for each digit.
Fortuna
Fortuna was created by a group of interns at Publicis Sapient, summer 2023. It is a brokerage application that simulates different types of market orders based on real-time stock data. The authentication layer is handled by Auth0, the back end is a Flask server, and the front end is made with React. I led the front end team and created the charts and tables seen on the application dashboard.
Quizify
Quizify is a chatGPT-powered multiple-choice quiz generation application during my time at Publicis Sapient as a 4-day website development sprint. The app uses React for the front-end, Express for the back-end, and Firebase to store high-scores.
Vari-Wordle
Vari-Wordle is a New York Times Wordle clone where the length of the generated word can span from 4 to 8 characters. This variation expands upon the challenge of the game by requiring a different strategy to win depending on the length of the given word. This was my first full-stack project and helped me refine my React skills.